Introduction to Artificial Intelligence

Artificial intelligence is now one of the most widely discussed, widely used, and misunderstood technologies.
We encounter its applications in search engines, recommendation systems, digital assistants, translation software, image and text generation, medicine, transportation, finance, and countless other fields. Yet, precisely because the term is used so frequently, its actual meaning is not always clear.

What makes a system “intelligent”?
What is the difference between a conventional program and an artificial intelligence system?
Does all artificial intelligence rely on machine learning?
And how did we move from early rule-based systems to today’s models capable of learning from data?
This book was written to answer these questions progressively and rigorously, building a comprehensive understanding of artificial intelligence before exploring its specific approaches.
Our starting point will be the very meaning of artificial intelligence.

There is, in fact, no single technique called AI. The term encompasses different approaches, developed at different points in history and based on different ideas about how a machine can perceive information, represent problems, make decisions, and produce behaviors appropriate to a given objective.
To truly understand this field, it is therefore useful to begin not with the latest tools, but with its conceptual foundations.
We will first examine what characterizes an artificial intelligence system.

An AI system does not operate in isolation. It receives information from an environment, represents it internally, processes it according to a particular model, and produces actions or outputs that may modify the environment itself. Sensors, data, representations, algorithms, models, objectives, and actions are therefore different components of a single process.
This perspective allows us to understand artificial intelligence not simply as software, but as a system that perceives, processes, and acts.

We will then explore what we can call classical artificial intelligence.
Before the widespread adoption of modern machine learning, many intelligent systems were built using rules, logic, search through solution spaces, knowledge representation, and planning. In these systems, much of the intelligence was explicitly designed by humans.
Studying these approaches is important because it helps us understand that artificial intelligence is not synonymous with machine learning.

A system can exhibit complex behaviors without learning from data, relying instead on strategies directly designed by its developers.
From there, we will move on to modern artificial intelligence, characterized above all by the growing importance of machine learning.
In this approach, the system’s behavior is not entirely defined through explicit instructions. Instead, the system uses data and examples to identify patterns and construct models capable of generalizing to new situations.
This shift profoundly changes the way intelligent systems are designed.

In the traditional programming paradigm, the programmer directly specifies the rules required to transform an input into an output. In machine learning, by contrast, some of these rules are inferred automatically from data through a learning process.
Understanding this difference is essential for interpreting much of contemporary artificial intelligence.
However, the purpose of this book is not simply to present a historical sequence of technologies.
Its goal is to develop a mental model of artificial intelligence.

To understand the elements that make up an intelligent system. To distinguish a traditional algorithm from a model that learns. To understand the role of data. To recognize the difference between explicitly designed rules and learned behaviors. To understand why different approaches can be used to solve different problems.
This perspective also helps us avoid two opposing but equally misleading interpretations.
On the one hand, viewing artificial intelligence as some kind of mysterious entity capable of “thinking” independently. On the other, reducing it to merely a new category of software.

Artificial intelligence is instead a multifaceted field, encompassing models, algorithms, data, representations, and strategies designed to produce behaviors that, in specific contexts, can be considered intelligent.
This book therefore aims to provide the foundations needed to explore more advanced topics, including machine learning, neural networks, generative models, agentic systems, and many of the technologies that are currently transforming the relationship between humans and digital systems.
Because before learning how to use artificial intelligence, it is useful to understand what we are actually referring to when we speak of artificial intelligence.